Skip to content
Kwun Chen · Data Science
Kwun Chen

Data Science + Statistics at UC Berkeley. I build pipelines that make noisy data usable.

experience.log

Experience

Data engineering, analytics and research work, newest first. Open-source work is at the bottom.

Anyscale

Anyscale

Data Engineering Consultant

Sep 2026 – Present

  • Building a config-driven Polars pipeline for an autonomous RC car platform. It logs multi-sensor data, cleans it, and loads it as analysis-ready data for GPU-based model training.
  • Polars
  • Python
  • Data pipelines
Air Combat Command (U.S. Air Force)

Air Combat Command (U.S. Air Force)

Data Science & Automation Intern

May 2026 – Aug 2026

  • Built an ELT pipeline and star schema that connected Palantir Envision with 500+ shared-drive folders and fed a weekly-refreshed Power BI dashboard. ACC’s colonels and brigadier generals used it for readiness insights.
  • Built an ingestion and validation pipeline that ran 4,000+ personnel records through a multi-criteria decision-tree eligibility engine. Checked against manual review, it reached 90% automated accuracy and saved 5+ hours of manual work a week.
  • Designed a unified data model for senior enlisted personnel and proposed a talent-identification dashboard that combines readiness, PME and performance data.
  • ELT
  • Star schema
  • Power BI
  • Palantir
Sawyer

Sawyer

Product & Research Consultant

Feb 2026 – May 2026

  • Collected and analyzed 500+ customer reviews from 100+ websites in one week, coding the ergonomics feedback into a structured dataset for a consumer product prototype.
  • Turned the analysis into findings that non-technical product stakeholders could act on. It showed which features customers responded to most and shaped the prototype’s direction.
  • Took part in a cross-functional review that narrowed 4 candidate prototype designs down to 1, using feedback from stakeholders outside the product team.
  • Research
  • Qualitative coding
  • Product
Air Force ROTC Detachment 085

Air Force ROTC Detachment 085

Lead Engineer

Oct 2025 – Present

  • Trained scikit-learn models on cadet performance data to find training gaps, and used the results to inform the detachment’s training decisions.
  • Leading a team that deployed an LLM-powered chatbot on Cloudflare Workers, which gives 100+ cadets real-time answers.
  • scikit-learn
  • LLMs
  • Cloudflare Workers
Open source

Open-source Roblox systems

Seven server-authoritative Luau packages built on a modular client-server architecture. They share a state-management service that settles conflicting writes when packages are installed independently of each other.

  • Luau
  • Roblox